The Mini Nutritional Assessment (MNA), developed in Europe in 1994, is presently considered to be a standard method for assessing the nutritional status of the elderly in western countries. In Japan, however, the MNA is not yet accepted as a reliable method for assessing the nutritional status of the Japanese super-elderly on the grounds that over 85% of Japanese elderly are simply included in at risk or malnourished groups, and that the MNA alone is not useful for selecting the poorest nourished among them. In this study, we investigated the applicability of the MNA to the Japanese super-elderly (mean age=87 years) by dividing the MNA at risk and malnourished groups into four subgroups using both the MNA scores alone and also in combination with serum albumin level (cut-off point, 3.5g/dL). Our results indicated that the MNA was also applicable to the Japanese super-elderly. When at risk or malnourished subjects were classified into four subgroups according to MNA scores in combination with the serum albumin level, it predicted the final outcome of institutionalized super-elderly and was helpful for selecting those requiring earlier nutritional interventions.